AI Summary & Key Takeaways
AI InsightHowever, the distance between centroids of roads and Hausdorff distances were more satisfactory.
📌 3 Key Takeaways
Abstract
Decision makers and researchers need datasets from different sources to analyze, combine, or create new spatial datasets. The same entity may be represented with different geometries, topologies, and attributes in different datasets due to differences in production, such as projection, scale, accuracy, purpose, and date. The geometries, topologies, and attributes of objects are often used when combining and integrating the datasets from different sources. Matching spatial datasets is one of the most important phases of data integration. Many algorithms have been developed to match datasets using several parameters inspired by geometric, topological, and attribute similarities. They generally find the similarities between objects in different datasets and create relations between each object in order to analyze, combine, update, and transfer data. The differences in geometries, topologies, and attributes make the matching process difficult. The research problem is the critical selection of similarity parameters to ensure the satisfactory matching results. The scope of this paper was limited with distance metrics. In this study, it was aimed to determine the suitable distance metrics measured from point to point and from point to line, which are widely used as parameters in road matching. Two road datasets in different databases were automatically matched using these metrics by employing a plugin of an open desktop software. Automatic matching results were compared to manual matching results to determine the success of each matching process. Consequently, it was shown that none of these metrics for road matching was adequate on its own. However, the distance between centroids of roads and Hausdorff distances were more satisfactory.